Recent empirical and modeling research has focused on thesemantic fluency task because it is informative about seman-tic memory. An interesting interplay arises between the rich-ness of representations in semantic memory and the complex-ity of algorithms required to process it. It has remained anopen question whether representations of words and their re-lations learned from language use can enable a simple searchalgorithm to mimic the observed behavior in the fluency task.Here we show that it is plausible to learn rich representationsfrom naturalistic data for which a very simple search algorithm(a random walk) can replicate the human patterns. We sug-gest that explicitly structuring knowledge about words into asemantic network plays a crucial role in modeling human be-havior in memory search and retrieval; moreover, this is thecase across a range of semantic information sources.